Query logs

Query logs is the deep-dive surface over system.query_log — every execution the cluster recorded, rolled up and charted five different ways. It lives at /monitoring/logs and requires the logs:view permission.

Five sub-views

The tab strip switches between five sub-views. All of them honor the shared time-range picker (15 m / 1 h / 6 h / 24 h presets plus a Grafana-style drill-down calendar from day → month → year in one popover).

Queries

Every execution as a dense, sortable table:

• SQL-keyword tooltip preview on hover — see the statement without opening the row
• Memory-pressure flags: a flame icon at ≥ 25 % cluster RAM, a triangle at ≥ 10 %
• Row checkboxes for side-by-side Compare
• Expanded-row drill-downs for Profile events and Views triggered (system.query_views_log)

Patterns

normalizeQuery() rollup with cumulative cost — Runs, Avg duration, Total duration, Max memory, Read rows, Read bytes. Default sort is Total duration DESC, which surfaces the patterns hogging the most wall-clock time across all repetitions, not just single slow queries.

By table

arrayJoin(tables) per-table rollup with arrayFilter push-down, so busy clusters return in roughly one second. Use it to answer "which table drives the load?".

By Redash

Groups every Redash-originated query by the query_id embedded in its leading SQL comment (/* … query_id: NNNN … */), mapping cluster load back to saved dashboard queries. Shows Runs, Min/Avg/Max/Total duration, Min/Max memory, read rows and read bytes — all sortable.

Histogram

Distribution of duration, memory, read rows and read bytes across the active window, with p50 / p95 / p99 chips. p99 is amber-tinted to flag the tail.

Timeline chart

Alongside the sub-views, the query timeline chart renders stacked bar, stacked area and line variants grouped per query_kind. It is the fastest way to spot a load spike before drilling into the tables.

Chouse AI integration

With the ai:optimize permission, a query-log row exposes Optimize with Chouse AI:

• An optimized rewrite that produces the same result
• A before → after EXPLAIN estimate as proof
• One click to Open in Explorer to run the rewritten query

The AI is read-only and advisory — review the rewrite before running it. See Chouse AI in-tab.

Underlying data